Update README.md
Browse files
README.md
CHANGED
@@ -11,4 +11,34 @@ tags:
|
|
11 |
pretty_name: MoleculeNet ToxCast
|
12 |
size_categories:
|
13 |
- 1K<n<10K
|
14 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
pretty_name: MoleculeNet ToxCast
|
12 |
size_categories:
|
13 |
- 1K<n<10K
|
14 |
+
---
|
15 |
+
|
16 |
+
# MoleculeNet ToxCast
|
17 |
+
|
18 |
+
ToxCast dataset [[1]](#1), part of MoleculeNet [[2]](#2) benchmark. It is intended to be used through
|
19 |
+
[scikit-fingerprints](https://github.com/scikit-fingerprints/scikit-fingerprints) library.
|
20 |
+
|
21 |
+
The task is to predict 617 toxicity targets from a large library of compounds based on in vitro high-throughput screening. All tasks are binary.
|
22 |
+
|
23 |
+
Note that targets have missing values. Algorithms should be evaluated only on present labels. For training data, you may want to impute them, e.g. with zeros.
|
24 |
+
|
25 |
+
| **Characteristic** | **Description** |
|
26 |
+
|:------------------:|:------------------------:|
|
27 |
+
| Tasks | 617 |
|
28 |
+
| Task type | multitask classification |
|
29 |
+
| Total samples | 8576 |
|
30 |
+
| Recommended split | scaffold |
|
31 |
+
| Recommended metric | AUROC |
|
32 |
+
|
33 |
+
## References
|
34 |
+
<a id="1">[1]</a>
|
35 |
+
Ann M. Richard et al.
|
36 |
+
"ToxCast Chemical Landscape: Paving the Road to 21st Century Toxicology"
|
37 |
+
Chem. Res. Toxicol. 2016, 29, 8, 1225–1251
|
38 |
+
https://pubs.acs.org/doi/10.1021/acs.chemrestox.6b00135>
|
39 |
+
|
40 |
+
<a id="2">[2]</a>
|
41 |
+
Wu, Zhenqin, et al.
|
42 |
+
"MoleculeNet: a benchmark for molecular machine learning."
|
43 |
+
Chemical Science 9.2 (2018): 513-530
|
44 |
+
https://pubs.rsc.org/en/content/articlelanding/2018/sc/c7sc02664a
|